BNL-68725

AEROSOL CHEMICAL CHARACTERIZATION ON BOARD THE DOE G1 AIRCRAFT USING A PARTICLE-INTO-LIQUID-SAMPLER DURING THE TEXAQS 2000 EXPERIMENT

Y.-N. Lee, Z. Song, Y. Liu, P. Daum, R. Weber, D. Orsini, N. Laulainen, J. Hubbe, and V. Morris

Atmospheric Sciences Division Brookhaven National Laboratory Upton, NY 11973-5000

October 2001

For presentation at the American Meteorological Society 82nd Annual Meeting, Orlando, FL January, 13-17, 2002

Research by BNL investigators was performed under the auspices of the U.S. Department of Energy under Contract No. DE-AC02-98CH10886. 10.13 AEROSOL CHEMICAL CHARACTERIZATION ON BOARD THE DOE G1 AIRCRAFT USING A PARTICLE-INTO-LIQUID-SAMPLER DURING THE TEXAQS 2000 EXPERIMENT

Y.-N. Lee*, Z. Song*, Y. Liu*, P. Daum*, R. Weber¹, D. Orsini¹, N. Laulainen**, J. Hubbe**, V. Morris** *Brookhaven National Laboratory, Upton, NY, ¹Georgia Institute of Technology, Atlanta, GA, **Pacific Northwest National Laboratory, Richland, WA

The PILS was constructed based on the design that was 1. INTRODUCTION used at the 1999 Atlanta EPA Supersite Experiment Knowledge of aerosol chemical composition is key (Weber et al., 2001), and was deployed on an aircraft to understanding a number of properties of ambient for the first time during TexAQS 2000. A schematic aerosol particles including sources, size/number diagram of the aircraft system is shown in Figure 1. The distribution, chemical evolution, optical properties and principle of the PILS technique involves activating the human health effects. Although filter based techniques particles under supersaturated conditions achieved by have been widely used to determine aerosol chemical mixing steam with the sample air and collecting the constituents, they generally cannot provide sufficiently resulting droplets using an impactor with a 50% cutoff at fast time resolution needed to investigate sources and ~1 mm. The liquid sample collected on the impactor chemical evolution that effect aerosol chemical, size surface is washed off with a steady stream of carrier and number changes. In order to gain an ability to flow and delivered to two sample loops (150 mL each) describe and predict the life cycles of ambient aerosols for separate anion and cation analysis using two IC’s. as a basis for ambient air quality control, fast and The samples were injected to the IC for analysis every 3 sensitive determination of the aerosol chemical min and each sample represented a time average of ~2 + + composition must be made available. To help to min. The following species were quantified: Na , NH4 , + 2+ - 2- achieve this goal, we deployed a newly developed K , Ca , NO3 , and SO4 . The limit of detection was technique, referred to as PILS (particle-into-liquid- ~0.1 ug m-3 for all the species at a limit of detection of sampler), on the DOE G1 aircraft during the 2000 ~0.1 mM of the IC and a sample air flow rate of 5.0 L Texas Air Quality Study (TexAQS 2000) to min-1. characterize the major ionic species of aerosol particles with aerodynamic size smaller than 2.5 mm Ambient air in (PM 2.5). The results obtained are examined in the Window Blank context of other simultaneously collected data for insights into the measurement capability of the PILS To pump, MOUDI 25 L/min system. size cut: 2.5 µm Denuder Na2CO3 2. EXPERIMENTAL SECTION Steam PC Generator Denuder TexAQS 2000 was a major field experiment Citric acid conducted to investigate photochemical O3 production Growth Cation IC in the greater Houston region where frequent O3 H2O in, Chamber exceedance over the 1 hr-120 ppbv NAAQS was 0.4 mL/min Impactor Waste observed. Because the Houston area has a highly Steam Heater Anion IC Controller To pump, concentrated petroleum and chemical industry, 5.0 L/min investigating the role industrial emissions play, in Carrier flow in, addition to that of urban emission, in O3 production was 0.20 mL/min a major focus of TexAQS 2000. This field program Figure 1. Schematic diagram of the PILS-IC system was led principally by the Texas Natural Resources deployed on the G1 aircraft. Conservation Commission, and a large number of research groups from universities, government and the The inlet to the PILS system consisted of a diffusor private sector had participated. Both ground cone (7 degree taper) with a 2.5 mm diameter nozzle for measurement sites and instrumented aircraft were near isokinetic sampling at the G1 cruising speed (~200 used to characterize the air chemistry. k hr-1), a MOUDI impactor for a 2.5 mm size cut, and two glass annular denuders in series to remove gaseous A PILS coupled to ion chromatography (IC) HNO and NH . While the total flow into the system analysis was deployed on board the DOE-G1 aircraft 3 3 was 30 L min-1 appropriate for the inlet and the MOUDI during TexAQS 2000 to determine the ionic impactor, 25 L min-1 was dumped after the MOUDI and components of aerosol particles. 5 L min-1 was sampled by the PILS (Fig. 1). *Corresponding author address: Y.-N. Lee, The G1 aircraft had a wide range of instrumentation Atmospheric Sciences Division, Brookhaven National for gas and aerosol characterization. With regard to Laboratory, Upton, NY 11973; email: [email protected] aerosols the most relevant instruments included a pcasp (passive cavity aerosol spectrometer, Particle as dates with a and b denoting morning and afternoon Measurement System, Boulder, CO) which determines flights, respectively. the size/number distribution of particles in 15 size bins within the range 0.1 to ~3 mm, and a PSAP (particle 16 soot absorption photometer, Radiance Research, Seattle, WA) which measures aerosol optical absorption due primarily to black carbon. 12 A total of 18 research flights were conducted between 8/19/00 and 9/12/00, of which PILS data were collected on 16. On flight days, typically two flights 8 were conducted, taking off at ~9:30 am and ~2:00 pm CDT, each lasting for ~2 hr. Although the flight tracks were slightly different between the AM and PM flights, 4 they cover the greater Houston region over the perimeter, as well as the downtown area and the ship channel area where the petrochemical industry 0 facilities are concentrated. The only flight that deviated from this pattern was the 0910a flight for a power plant plume study northeast of the Houston area (Springston Figure 3. Box plots of total aerosol ion mass et al., 2002). The altitude was maintained between concentrations determined on each flight. 400 to 600 m except during spirals to determine the boundary layer height. An important aerosol property is its acidity, or the extent of neutralization of sulfuric acid present in the 3. RESULTS AND DISCUSSION aerosol. One measure used to gauge this property is the examination of the ratio (j) of the principal + 2- The PILS-IC system measured the inorganic neutralizing agent, NH4 , to the sum of 2xSO4 and + + + - aerosol ionic components, namely, Na , NH4 , K, NO3 . A value of unity indicates complete neutralization 2+ - 2- + + 2- Ca , NO3 , and SO4 . Although Na , NH4 and SO4 of nitric and sulfuric acids by ammonia. Departure from were present in all samples, Na+ was present at a unity suggests the presence of other acids, e.g., organic much lower level (< 0.3 µg m-3) than the other two. The acids, if j>1 or the presence of free strong acids if j<1. + median and maximum concentrations of NH4 and The observed j showed a fairly wide distribution, with 2- -3 SO4 were 1.6 and 5.0, and 1.8 and 9.4 µg m , the mode being between 1 and 2, suggesting additional - -3 + respectively. NO3 was mostly <0.2 µg m , but with anions to associate the excess NH4 . In Figure 4 we 2- 2- occasional excursion reaching as high as half of SO4 . show box plots of the j values as a function of [SO ] in + 2+ 4 K and Ca were seen in only ~15% of the samples, 4 quartiles. Not surprisingly, j exhibited the highest but K+ reached as high as 6 mg m-3. The distributions

of these species are shown as frequency plots in ) 30 6 5 10 2- Figure 2. In addition, the distribution of the total mass 4 + 2*SO

- 4 8 10 3 /(NO

+ 20 4 4

3 6 8 SO 4

6 2 4 -3 10 2

NH or Molar Ratio, NH 4 -3 4 1 2 Conc., ug m ], ug m 2- NO 4 K 3 2 [SO 0 0 0 0 SO4 Ratio SO4 Ratio SO4 Ratio SO4 Ratio

0 Na Ca + Figure 4. Box plots of the ratio of [NH4 ] to the sum of - 2- 2- -2 [NO ] + 2*[SO ] for each quartile of SO mass .01 .1 1 5 10 20 30 50 70 80 90 95 99 99.9 99.99 3 4 4 Percent concentration. 2- values at the lowest SO4 concentration and decreased 2- Figure 2. Probability distribution of the aerosol with increasing SO4 mass loading. However, even for ionic species. th 2- the 4 [SO4 ] quartile, j ranged between 0.5 and 2 with of the ionic species are shown as box plots for each of a median value of 1.2, strongly suggests the presence the flights (Figure 3). Note that the flights are labeled of other anions, presumably organic in nature. One of + the possible artificial causes of an elevation of NH4 2000), we have corrected the pcasp data using a RI of concentrations, and therefore j, is an depleted 1.40. denuder that allowed gaseous NH3 to be collected and + The total aerosol mass was estimated with an initial counted as aerosol NH . To examine this possibility, 4 assumption of a density of 1.7 g cm-3 appropriate for we eliminated the data between 8/26 and 9/1 during (NH ) SO 2-. However, if the ion mass fraction was which j was in general the highest. In this subset of 4 2 4 found to account for £70 % of the initial estimated total data, the j values ranged between 0.5 and 6 with a mass, then a reduced density was applied which median value of 1.7. Consequently the overall picture assumed a linear combination of 1.7 for the inorganic concerning the excess NH + comparing to NO - and 4 3 fraction and 1.0 for the rest which conceivably is due to SO 2- is not changed. 4 organic components. Although an iterative process can The activation and collection efficiency of the PILS be taken to reach a final total mass, the fact that the has been characterized in the laboratory using total aerosol volume also has a sizable uncertainty (NH4)2SO4 and tested in the field during the 1999 EPA owing to the assumed RI makes it unwarranted to adjust Atlanta Supersite Experiment (Weber et al., 2001). the density beyond the second step. We surmise that The laboratory results showed that particles with a the estimated aerosol mass has a 15% uncertainty. The diameter greater than 100 nm are quantitatively total aerosol mass estimated based on this procedure is collected. In addition, the field results showed that referred to as pcasp mass. aerosol mass was nearly quantitatively captured based The pcasp mass in general correlated well with the on TEOM (Tapered Element Oscillation Microbalance, total ion mass and the correlation coefficients ranged Rupprecht and Patashnick, Albany, NY) measurement. between 0.4 to 0.8, except for flights 0831a, 0831b, For the current aircraft study, it is also desirable to 0907a, and 0910b. Additional effort will be applied to compare the PILS-IC data with the total aerosol mass these data to understand the sources of the poor for the purpose of examining the mass contributions of correlation. For those that exhibit a good correlation, we the inorganic ionic species. In the absence of a direct display as examples two flights, 0906b and 0912a, aerosol mass measurement on the G1, we estimated where, respectively, nearly the highest and lowest the total aerosol mass using the number/size aerosol mass loadings were observed during the distribution of the accumulation mode particles mission (Fig. 5). In these analyses, the pcasp mass determined using the pcasp with an assumed density. were estimated using 1.25 and 1.7 as density for 0906p It should be noted that the raw pcasp data were and 0911a, respectively. determined with the instrument calibrated using The relatively strong correlation between the pcasp polystyrene latex spheres that have a refractive index mass and the total ion mass, especially for those flights (RI) of ~1.58. Because this RI value is not where a low aerosol mass loading was found (and the representative of ambient aerosols (Liu and Daum,

50 15

0906b Figure 5. Left column: 40 -3 time series plots of the

-3 10 pcasp mass and the 30 total ion mass on 9/6/00 and 9/12/00, when the 20 highest and the lowest 5

Conc., ug m aerosol mass loading, 10 respectively, were Total Ion Mass, ug m y = 4.04 + 0.203x R 2= 0.61 observed. Right 0 0 column: correlation 20:00:00 21:00:00 22:00:00 0 10 20 30 40 50 between the pcasp and 15 8 total ion mass. 0912a -3 6

-3 10

4

5

Conc., ug m 2 Total Ion Mass, ug m y = 1.13 + 0.489x R 2= 0.81 0 0 15:00:00 16:00:00 17:00:00 0 5 10 15 Time of Day, UTC Total Aerosol Mass, ug m-3 ionic components accounted for a major portion of the carbon mass loading was derived from the PSAP data total aerosol mass), suggests that the PILS using a specific absorption coefficient of 10 m2 g-1, albeit measurements are indeed reasonable. The difference it was well recognized this coefficient is variable between the pcasp mass and the total ion mass may depending on the environment (Liousse et al, 1993). therefore be attributed to organic constituents of the Figure 6 shows a fairly strong correlation between the aerosol particles. This approach of inference will be mass difference between pcasp and total ion and the useful to complement (and setting an upper bound limit) black carbon mass, supporting that the mass difference the existing techniques (e.g., the EC/OC technique by is at least dominated by organic components. However, Chow et al., 1993) to gain an improved understanding of it is interesting to note that the slopes are significantly the contribution of organics to aerosol mass loading different between 0906b and 0912a, the latter being which is critically needed (Huebert and Charlson, 2000). fairly close to that observed by Turpin and Huntzicker (1995) for a primary emission dominated situation. The In this connection, we examine the relationship much higher slope of 0906b may strongly suggest a between the black carbon and the difference between case of secondary organic aerosol formation. the pcasp mass and the total ion mass, which we tentatively term as the organic fraction, shown for the Acknowledgements. two flights, 0906b and 0912a (Figure 6). Good correlation has been observed between organic carbon This research was supported by the Atmospheric Chemistry Program within the DOE/OBER and performed under contract DE-AC02-98CH10886. The 40 authors wish to thank the pilots and crew of the DOE G1 0906b aircraft for their professional assistance.

-3 30 References: Chow, J.C., J.G. Watson, L.C. Pritchett, W.R. Pierson, C.A. Frazier and R.G. Purcell. 1993: The DRI 20 thermal/optical relectance carbon analysis system: description, evaluation and applications in US air quality studies. Atmos. Environ., 27A, 1185-1201.

Mass Difference, ug m 10 Hueber, B.J. and R.J. Charlson. 2000: Uncertainties in data on organic aerosols. Tellus, 52B, 1249-1255. y = 2.78 + 8.29x R2= 0.78 Liousse, C., H. Cachier, and S.G. Jennings. 1993: 0 Optical and thermal measurements of black carbon 0 1 2 3 4 -3 aerosol content in different environments: variation of BC Mass, ug m the specific attenuation cross section, sigma (s). Atmos. Environ., 27A, 1203-1211.

5 Liu, Y. and P.H. Daum, 2000: The effect of refractive index on size distributions and light scattering 0911a coefficients derived from optical particle counters. J. 4 Aerosol Sci., 31, 945-957. -3 Springston, S.R., L.I. Kleinman, F. Brechtel, Y.-N. Lee, 3 L.J. Nunnermacker and J. Weinstein-Lloyd. 2002: Chemical evolution of a power-plant plume. Abstract to be presented at the 82nd AMS Annual Meeting, 2 Orlando, FL. January 13-17, 2002. Paper no. P1.29. Turpin, B. and J.J. Huntzicker. 1995: Identification of Mass Difference, ug m secondary organic aerosol episodes and quantitation of 1 primary and secondary organic aerosol concentrations y = 0.35 + 3.20x R 2= 0.50 during SCAQS. Atmos. Environ., 29, 3527-3544. 0 0 0.5 1 1.5 Weber, R.J., D. Orsini, Y. Duan, Y.-N. Lee, P. J. Klotz, BC Mass, ug m -3 and F. Brechtel. 2001: A particle-into-liquid-collector for rapid measurement of aerosol bulk chemical Figure 6. Correlation between the organic fraction composition. Aerosol Science and Technology, 35, (see text) and the black carbon mass concentration. 718-727. and elemental carbon when primary emission was important (e.g., Turpin and Huntzicker, 1995), The black